sciGPGPU is a Scilab Toolbox to allow interaction between Scilab and CUDA/OpenCL.

---------------------- Linux ---------------------- 

	Requirements :
		- CUDA Toolkit >= 3.2
		- scilab >= 5.2.1 (could work with precedent version, but this was not tested)
		- boost libraries >= 1.4.0

	How to build from tar.gz file :
		- Unzip the tar.gz in your SCI/contrib path.
		- Launch scilab and 'cd' to SCI/contrib/sciGPGPU-* path.
		- exec builder.sce
		- Restart your scilab, now sciGPGPU should be available in the toolbox menu.

	Please modify the nvccdir.txt to write the nvcc executable.(by default, /usr/local/cuda/bin/nvcc)
	Please modify the cudaLib.txt to write the library path. One path by line. (by default, /usr/local/cuda/lib)
	Please modify the cudaInc.txt to write the include path. One path by line. (by default, /usr/local/cuda/inc)


---------------------- Windows ---------------------- 

	Requirements :
		- CUDA Toolkit >= 3.2
		- scilab >= 5.2.1 (could work with precedent version, but this was not tested)
		- Visual Studio 2009

	How to build from .zip file :
		- Unzip the .zip in your SCI/contrib path.
		- Launch scilab and 'cd' to SCI/contrib/sciGPGPU-* path.
		- exec builder.sce
		- Restart your scilab, now sciGPGPU should be available in the toolbox menu.


---------------------- All system ----------------------

	How to build from git :
		- Clone the repository with "git clone git@git.forge.scilab.org:sciCuda.git".
		- cd sciCuda/sciGPGPU
		- Launch scilab
		- 'exec builder.sce'
		- 'exec loader.sce'

	How to choose the compilation mode:

		At the top of the builder.sce file there are two variable :
			-WITH_CUDA, allow the toolbox to use Cuda. 
			-WITH_OPENCL, allow the toolbox to use OpenCL.
		The two variables can be true, then you will be able to switch between Cuda and OpenCL after the toolbox building.
